Level 3 · How it worksThe gold rush#190

Mining difficulty explained: the creek gets harder every time the crowd grows

Mining difficulty sets how hard it is to find a valid Bitcoin block. Every 2,016 blocks, every node recalculates it from how long those blocks took, aiming at ten minutes per block, and it can change by at most a factor of four at once.

In a gold rush, the first men at a creek found nuggets lying in the shallows. Then the crowd arrived. More pans in the same water meant less gold for each of them, and the easy finds were gone.

Bitcoin builds that effect in on purpose. When more miners join, blocks start coming faster than planned. So every two weeks or so, the network makes the search harder. When miners leave, it makes it easier.

That knob is the mining difficulty. Nobody turns it by hand. Every node works it out from the blockchain's own clock, and the result keeps Bitcoin's pace close to one block every ten minutes, however many pickaxes show up.

What it is

Difficulty is a measure of how hard it is to find a block hash below the network's target. A lower target means a higher difficulty. Bitcoin Core sets the goal at 600 seconds per block and recalculates over a period of "14 * 24 * 60 * 60" seconds, two weeks (chainparams.cpp). Two weeks of ten-minute blocks is 2,016 blocks, so the difficulty changes once every 2,016 blocks and stays fixed in between.

The white paper describes it as a moving target: "If they're generated too fast, the difficulty increases" (Nakamoto, 2008).

How it works: two real retargets in autumn 2026

The rule in Bitcoin Core's pow.cpp has three steps (pow.cpp):

  1. At every height divisible by 2,016, look back at the period just finished. The code's comment reads: "Go back by what we want to be 14 days worth of blocks".
  2. Measure the time between the period's first and last block, using the timestamps in their headers.
  3. Multiply the old target by that time and divide by two weeks. Under the comment "Limit adjustment step", the measured time is clamped between a quarter of two weeks (3.5 days) and four times two weeks (56 days). So difficulty can rise or fall by at most a factor of four at once.

Here are two consecutive retargets, read from the mempool.space explorer on 19 November 2026 (mempool.space).

Retarget at block 973,728. The period ran from block 971,712, mined on 16 October 2026 at 12:37:31 UTC (mempool.space), to block 973,727, mined on 29 October 2026 at 18:59:30 UTC (mempool.space). That is 1,146,119 seconds, about 13.27 days instead of 14. Blocks came every 9.5 minutes on average.

  • Time ratio: 1,146,119 / 1,209,600 = 0.9475.
  • Old difficulty: 140,577,680,051,920. New difficulty: 140,577,680,051,920 / 0.9475 = 148,364,561,003,867.
  • Change: up 5.54%.

Retarget at block 975,744. The next period ran from block 973,728 (29 October 2026, 19:14:58 UTC) to block 975,743 (14 November 2026, 04:59:43 UTC) (mempool.space). That is 1,331,085 seconds, about 15.41 days. Blocks came every 11 minutes on average.

  • Time ratio: 1,331,085 / 1,209,600 = 1.1004.
  • New difficulty: 148,364,561,003,867 / 1.1004 = 134,823,837,323,273.
  • Change: down 9.13%.

We recomputed both from the block timestamps and the compact target field ("bits") in the block headers, with the same integer arithmetic Bitcoin uses. Both results match the target written into blocks 973,728 and 975,744 to the last digit. Up one fortnight, down the next: the knob works in both directions, and neither move came close to the factor-four limit.

Why it matters

  • A steady clock. New coins arrive at a pace set by block height, not by how many miners show up. More machines don't print more bitcoin. They just make the creek harder to pan.
  • Miners carry the risk. When difficulty rises, every machine earns less. When it falls, the miners who stayed get a bigger share. Business planning for miners depends on a number no one controls.
  • Nobody holds the knob. No company, foundation or vote sets the difficulty. A node that accepted a block with the wrong target would simply be out of step with the rest.
  • A two-week lag. The rule only reacts after the fact. If a large share of miners switched off at once, blocks would be slow until the next retarget. The factor-four cap limits how fast it can catch up.

Where people get it wrong

  • "Difficulty adjusts every two weeks." It adjusts every 2,016 blocks. If blocks are fast, that is sooner than two weeks. If they're slow, later.
  • "Higher difficulty means a stronger price." Difficulty follows the amount of mining, not the price. The saloon likes to read price signals into it. The code doesn't mention a price.
  • "Developers raise the difficulty." The rule is the same for every period and runs on every node. Changing it would take a change to Bitcoin's rules that nodes agree to run.
  • "Difficulty makes each block contain more work for users." Transactions are unaffected. Only the miners' search gets harder or easier.
  • "Difficulty can jump without limit." It can move at most by a factor of four per retarget. In the 26 retargets from November 2025 to November 2026, the biggest rise was 14.7% and the biggest fall 11.2% (mempool.space).

Check the next retarget yourself

Find the current block height on any block explorer. Divide by 2,016 and look at the remainder: that's how many blocks of the period have passed. Note the timestamp of the period's first block. Work out the average time per block since then and compare it with ten minutes. Faster than ten, the creek gets harder. Slower, easier.

That's the whole forecast. Anyone selling you a premium difficulty prediction service is selling a calculator, with a mark-up.

Frequently asked questions

What is Bitcoin mining difficulty?

Mining difficulty measures how hard it is to find a block whose hash is below the network's target. A higher difficulty means a lower target and more guesses needed on average. It is recalculated every 2,016 blocks.

How often does Bitcoin difficulty change?

Once every 2,016 blocks, which takes about two weeks if blocks arrive every ten minutes on average. Between those points the difficulty stays fixed.

How is the new difficulty calculated?

Every node measures how long the last 2,016 blocks took and scales the target by that time divided by two weeks. Faster than two weeks means a higher difficulty; slower means a lower one. On 14 November 2026, at block 975,744, slow blocks led to a 9.13% fall.

Is there a limit to how much difficulty can change?

Yes. Bitcoin Core clamps the measured time between 3.5 days and 56 days, so a single retarget can change difficulty by at most a factor of four in either direction.

Does difficulty affect transaction fees?

Not directly. Difficulty only changes how hard the miners' search is. Fees depend on how many people want space in the next blocks.

Sources

  1. Bitcoin Core source code, src/pow.cpp (GetNextWorkRequired, CalculateNextWorkRequired)
  2. Bitcoin Core source code, src/kernel/chainparams.cpp (nPowTargetTimespan, nPowTargetSpacing)
  3. Satoshi Nakamoto, Bitcoin: A peer-to-peer electronic cash system (2008), section 4
  4. mempool.space API, difficulty adjustments November 2025 to November 2026 (read 19 November 2026)
  5. mempool.space API, blocks 971,698 to 971,712 (16 October 2026)
  6. mempool.space API, blocks 973,714 to 973,728 (29 October 2026)
  7. mempool.space API, blocks 975,730 to 975,744 (14 November 2026)

Read next.